Cleanup text code

This Cl removes unused defines and methods. It also merges fx_arb into fx_arabic
as the arb methods are only used in the arabic file and they are always included
together.

Change-Id: I1e6f75d5cb5f5bcc8db3b54cb2a14d284da5b7b1
Reviewed-on: https://pdfium-review.googlesource.com/2899
Commit-Queue: dsinclair <dsinclair@chromium.org>
Reviewed-by: Tom Sepez <tsepez@chromium.org>
Reviewed-by: Nicolás Peña <npm@chromium.org>

9 files changed, 109 insertions, 176 deletions

diff --git a/BUILD.gn b/BUILD.gn
index 66a18bbc0c..8656ee7fb5 100644
--- a/BUILD.gn
+++ b/BUILD.gn
@@ -796,7 +796,6 @@ static_library("fxcrt") {
     sources += [
       "core/fxcrt/fx_arabic.cpp",
       "core/fxcrt/fx_arabic.h",
-      "core/fxcrt/fx_arb.h",
     ]
   }
 }
diff --git a/core/fxcrt/fx_arabic.cpp b/core/fxcrt/fx_arabic.cpp
index 108c6c19ac..29e6e437e8 100644
--- a/core/fxcrt/fx_arabic.cpp
+++ b/core/fxcrt/fx_arabic.cpp
@@ -286,17 +286,6 @@ FX_WCHAR FX_GetArabicFromShaddaTable(FX_WCHAR shadda) {
 namespace pdfium {
 namespace arabic {
 
-bool IsArabicChar(FX_WCHAR wch) {
-  uint32_t dwRet =
-      kTextLayoutCodeProperties[(uint16_t)wch] & FX_CHARTYPEBITSMASK;
-  return dwRet >= FX_CHARTYPE_ArabicAlef;
-}
-
-bool IsArabicFormChar(FX_WCHAR wch) {
-  return (kTextLayoutCodeProperties[(uint16_t)wch] & FX_CHARTYPEBITSMASK) ==
-         FX_CHARTYPE_ArabicForm;
-}
-
 FX_WCHAR GetFormChar(FX_WCHAR wch, FX_WCHAR prev, FX_WCHAR next) {
   CFX_Char c(wch, kTextLayoutCodeProperties[(uint16_t)wch]);
   CFX_Char p(prev, kTextLayoutCodeProperties[(uint16_t)prev]);
diff --git a/core/fxcrt/fx_arabic.h b/core/fxcrt/fx_arabic.h
index e7ba079f43..9f6ec50ef6 100644
--- a/core/fxcrt/fx_arabic.h
+++ b/core/fxcrt/fx_arabic.h
@@ -7,7 +7,10 @@
 #ifndef CORE_FXCRT_FX_ARABIC_H_
 #define CORE_FXCRT_FX_ARABIC_H_
 
-#include "core/fxcrt/fx_arb.h"
+#include <vector>
+
+#include "core/fxcrt/fx_system.h"
+#include "core/fxcrt/fx_ucd.h"
 
 #define FX_BIDIMAXLEVEL 61
 #define FX_BidiDirection(a) (FX_IsOdd(a) ? FX_BIDICLASS_R : FX_BIDICLASS_L)
@@ -17,9 +20,7 @@
 namespace pdfium {
 namespace arabic {
 
-bool IsArabicChar(FX_WCHAR wch);
-bool IsArabicFormChar(FX_WCHAR wch);
-FX_WCHAR GetFormChar(FX_WCHAR wch, FX_WCHAR prev = 0, FX_WCHAR next = 0);
+FX_WCHAR GetFormChar(FX_WCHAR wch, FX_WCHAR prev, FX_WCHAR next);
 FX_WCHAR GetFormChar(const CFX_Char* cur,
                      const CFX_Char* prev,
                      const CFX_Char* next);
@@ -46,47 +47,27 @@ int32_t FX_BidiResolveExplicit(int32_t iBaseLevel,
                                int32_t iNest = 0);
 
 enum FX_BIDIWEAKSTATE {
-  FX_BIDIWEAKSTATE_xa = 0,
-  FX_BIDIWEAKSTATE_xr,
-  FX_BIDIWEAKSTATE_xl,
-  FX_BIDIWEAKSTATE_ao,
-  FX_BIDIWEAKSTATE_ro,
-  FX_BIDIWEAKSTATE_lo,
-  FX_BIDIWEAKSTATE_rt,
-  FX_BIDIWEAKSTATE_lt,
-  FX_BIDIWEAKSTATE_cn,
-  FX_BIDIWEAKSTATE_ra,
-  FX_BIDIWEAKSTATE_re,
-  FX_BIDIWEAKSTATE_la,
-  FX_BIDIWEAKSTATE_le,
-  FX_BIDIWEAKSTATE_ac,
-  FX_BIDIWEAKSTATE_rc,
-  FX_BIDIWEAKSTATE_rs,
-  FX_BIDIWEAKSTATE_lc,
-  FX_BIDIWEAKSTATE_ls,
-  FX_BIDIWEAKSTATE_ret,
-  FX_BIDIWEAKSTATE_let,
+  FX_BWSxa = 0,
+  FX_BWSxr,
+  FX_BWSxl,
+  FX_BWSao,
+  FX_BWSro,
+  FX_BWSlo,
+  FX_BWSrt,
+  FX_BWSlt,
+  FX_BWScn,
+  FX_BWSra,
+  FX_BWSre,
+  FX_BWSla,
+  FX_BWSle,
+  FX_BWSac,
+  FX_BWSrc,
+  FX_BWSrs,
+  FX_BWSlc,
+  FX_BWSls,
+  FX_BWSret,
+  FX_BWSlet
 };
-#define FX_BWSxa FX_BIDIWEAKSTATE_xa
-#define FX_BWSxr FX_BIDIWEAKSTATE_xr
-#define FX_BWSxl FX_BIDIWEAKSTATE_xl
-#define FX_BWSao FX_BIDIWEAKSTATE_ao
-#define FX_BWSro FX_BIDIWEAKSTATE_ro
-#define FX_BWSlo FX_BIDIWEAKSTATE_lo
-#define FX_BWSrt FX_BIDIWEAKSTATE_rt
-#define FX_BWSlt FX_BIDIWEAKSTATE_lt
-#define FX_BWScn FX_BIDIWEAKSTATE_cn
-#define FX_BWSra FX_BIDIWEAKSTATE_ra
-#define FX_BWSre FX_BIDIWEAKSTATE_re
-#define FX_BWSla FX_BIDIWEAKSTATE_la
-#define FX_BWSle FX_BIDIWEAKSTATE_le
-#define FX_BWSac FX_BIDIWEAKSTATE_ac
-#define FX_BWSrc FX_BIDIWEAKSTATE_rc
-#define FX_BWSrs FX_BIDIWEAKSTATE_rs
-#define FX_BWSlc FX_BIDIWEAKSTATE_lc
-#define FX_BWSls FX_BIDIWEAKSTATE_ls
-#define FX_BWSret FX_BIDIWEAKSTATE_ret
-#define FX_BWSlet FX_BIDIWEAKSTATE_let
 
 enum FX_BIDIWEAKACTION {
   FX_BIDIWEAKACTION_IX = 0x100,
@@ -139,19 +120,14 @@ void FX_BidiResolveWeak(int32_t iBaseLevel,
                         CFX_ArrayTemplate<int32_t>& classes,
                         CFX_ArrayTemplate<int32_t>& levels);
 enum FX_BIDINEUTRALSTATE {
-  FX_BIDINEUTRALSTATE_r = 0,
-  FX_BIDINEUTRALSTATE_l,
-  FX_BIDINEUTRALSTATE_rn,
-  FX_BIDINEUTRALSTATE_ln,
-  FX_BIDINEUTRALSTATE_a,
-  FX_BIDINEUTRALSTATE_na,
+  FX_BNSr = 0,
+  FX_BNSl,
+  FX_BNSrn,
+  FX_BNSln,
+  FX_BNSa,
+  FX_BNSna
 };
-#define FX_BNSr FX_BIDINEUTRALSTATE_r
-#define FX_BNSl FX_BIDINEUTRALSTATE_l
-#define FX_BNSrn FX_BIDINEUTRALSTATE_rn
-#define FX_BNSln FX_BIDINEUTRALSTATE_ln
-#define FX_BNSa FX_BIDINEUTRALSTATE_a
-#define FX_BNSna FX_BIDINEUTRALSTATE_na
+
 enum FX_BIDINEUTRALACTION {
   FX_BIDINEUTRALACTION_nL = FX_BIDICLASS_L,
   FX_BIDINEUTRALACTION_En = (FX_BIDICLASS_AN << 4),
@@ -166,6 +142,35 @@ enum FX_BIDINEUTRALACTION {
 #define FX_BNALn FX_BIDINEUTRALACTION_Ln
 #define FX_BNAIn FX_BIDINEUTRALACTION_In
 #define FX_BNALnL FX_BIDINEUTRALACTION_LnL
+
+struct FX_ARBFORMTABLE {
+  uint16_t wIsolated;
+  uint16_t wFinal;
+  uint16_t wInitial;
+  uint16_t wMedial;
+};
+
+struct FX_ARAALEF {
+  uint16_t wAlef;
+  uint16_t wIsolated;
+};
+
+struct FX_ARASHADDA {
+  uint16_t wShadda;
+  uint16_t wIsolated;
+};
+
+const FX_ARBFORMTABLE* FX_GetArabicFormTable(FX_WCHAR unicode);
+FX_WCHAR FX_GetArabicFromAlefTable(FX_WCHAR alef);
+FX_WCHAR FX_GetArabicFromShaddaTable(FX_WCHAR shadda);
+
+void FX_BidiLine(std::vector<CFX_TxtChar>& chars,
+                 int32_t iCount,
+                 int32_t iBaseLevel = 0);
+void FX_BidiLine(std::vector<CFX_RTFChar>& chars,
+                 int32_t iCount,
+                 int32_t iBaseLevel = 0);
+
 int32_t FX_BidiGetDeferredNeutrals(int32_t iAction, int32_t iLevel);
 int32_t FX_BidiGetResolvedNeutrals(int32_t iAction);
 void FX_BidiResolveNeutrals(int32_t iBaseLevel,
diff --git a/core/fxcrt/fx_arb.h b/core/fxcrt/fx_arb.h
deleted file mode 100644
index d24197c7e5..0000000000
--- a/core/fxcrt/fx_arb.h
+++ /dev/null
@@ -1,51 +0,0 @@
-// Copyright 2014 PDFium Authors. All rights reserved.
-// Use of this source code is governed by a BSD-style license that can be
-// found in the LICENSE file.
-
-// Original code copyright 2014 Foxit Software Inc. http://www.foxitsoftware.com
-
-#ifndef CORE_FXCRT_FX_ARB_H_
-#define CORE_FXCRT_FX_ARB_H_
-
-#include <vector>
-
-#include "core/fxcrt/fx_system.h"
-#include "core/fxcrt/fx_ucd.h"
-
-struct FX_ARBFORMTABLE {
-  uint16_t wIsolated;
-  uint16_t wFinal;
-  uint16_t wInitial;
-  uint16_t wMedial;
-};
-
-struct FX_ARAALEF {
-  uint16_t wAlef;
-  uint16_t wIsolated;
-};
-
-struct FX_ARASHADDA {
-  uint16_t wShadda;
-  uint16_t wIsolated;
-};
-
-const FX_ARBFORMTABLE* FX_GetArabicFormTable(FX_WCHAR unicode);
-FX_WCHAR FX_GetArabicFromAlefTable(FX_WCHAR alef);
-FX_WCHAR FX_GetArabicFromShaddaTable(FX_WCHAR shadda);
-
-enum FX_ARBPOSITION {
-  FX_ARBPOSITION_Isolated = 0,
-  FX_ARBPOSITION_Final,
-  FX_ARBPOSITION_Initial,
-  FX_ARBPOSITION_Medial,
-};
-
-void FX_BidiLine(CFX_WideString& wsText, int32_t iBaseLevel = 0);
-void FX_BidiLine(std::vector<CFX_TxtChar>& chars,
-                 int32_t iCount,
-                 int32_t iBaseLevel = 0);
-void FX_BidiLine(std::vector<CFX_RTFChar>& chars,
-                 int32_t iCount,
-                 int32_t iBaseLevel = 0);
-
-#endif  // CORE_FXCRT_FX_ARB_H_
diff --git a/core/fxcrt/fx_ucd.h b/core/fxcrt/fx_ucd.h
index eb4bad097e..b07b387ece 100644
--- a/core/fxcrt/fx_ucd.h
+++ b/core/fxcrt/fx_ucd.h
@@ -49,46 +49,48 @@ uint32_t FX_GetUnicodeProperties(FX_WCHAR wch);
 FX_WCHAR FX_GetMirrorChar(FX_WCHAR wch, bool bRTL, bool bVertical);
 
 #ifdef PDF_ENABLE_XFA
+
+// As defined in http://www.unicode.org/reports/tr14/
 enum FX_CHARBREAKPROP {
-  FX_CBP_OP = 0,
-  FX_CBP_CL = 1,
-  FX_CBP_QU = 2,
-  FX_CBP_GL = 3,
-  FX_CBP_NS = 4,
-  FX_CBP_EX = 5,
-  FX_CBP_SY = 6,
-  FX_CBP_IS = 7,
-  FX_CBP_PR = 8,
-  FX_CBP_PO = 9,
-  FX_CBP_NU = 10,
-  FX_CBP_AL = 11,
-  FX_CBP_ID = 12,
-  FX_CBP_IN = 13,
-  FX_CBP_HY = 14,
-  FX_CBP_BA = 15,
-  FX_CBP_BB = 16,
-  FX_CBP_B2 = 17,
-  FX_CBP_ZW = 18,
-  FX_CBP_CM = 19,
-  FX_CBP_WJ = 20,
-  FX_CBP_H2 = 21,
-  FX_CBP_H3 = 22,
-  FX_CBP_JL = 23,
-  FX_CBP_JV = 24,
-  FX_CBP_JT = 25,
-
-  FX_CBP_BK = 26,
-  FX_CBP_CR = 27,
-  FX_CBP_LF = 28,
-  FX_CBP_NL = 29,
-  FX_CBP_SA = 30,
-  FX_CBP_SG = 31,
-  FX_CBP_CB = 32,
-  FX_CBP_XX = 33,
-  FX_CBP_AI = 34,
-  FX_CBP_SP = 35,
-  FX_CBP_TB = 37,
+  FX_CBP_OP = 0,   // Opening Punctuation
+  FX_CBP_CL = 1,   // Closing Punctuation
+  FX_CBP_QU = 2,   // Ambiguous Quotation
+  FX_CBP_GL = 3,   // Non-breaking ("Glue")
+  FX_CBP_NS = 4,   // Non Starter
+  FX_CBP_EX = 5,   // Exclamation/Interrogation
+  FX_CBP_SY = 6,   // Symbols Allowing Breaks
+  FX_CBP_IS = 7,   // Infix Separator (Numeric)
+  FX_CBP_PR = 8,   // Prefix (Numeric)
+  FX_CBP_PO = 9,   // Postfix (Numeric)
+  FX_CBP_NU = 10,  // Numeric
+  FX_CBP_AL = 11,  // Ordinary Alphabetic and Symbol Characters
+  FX_CBP_ID = 12,  // Ideographic
+  FX_CBP_IN = 13,  // Inseparable
+  FX_CBP_HY = 14,  // Hyphen
+  FX_CBP_BA = 15,  // Break Opportunity After
+  FX_CBP_BB = 16,  // Break Opportunity Before
+  FX_CBP_B2 = 17,  // Break Opportunity Before and After
+  FX_CBP_ZW = 18,  // Zero Width Space
+  FX_CBP_CM = 19,  // Attached Characters and Combining Marks
+  FX_CBP_WJ = 20,  // Word Joiner
+  FX_CBP_H2 = 21,  // Hangul LV Syllable
+  FX_CBP_H3 = 22,  // Hangul LVT Syllable
+  FX_CBP_JL = 23,  // Hangul Leading Jamo
+  FX_CBP_JV = 24,  // Hangul Vowel Jamo
+  FX_CBP_JT = 25,  // Hangul Trailing Jamo
+
+  FX_CBP_BK = 26,  // Mandatory Break
+  FX_CBP_CR = 27,  // Carriage Return
+  FX_CBP_LF = 28,  // Line Feed
+  FX_CBP_NL = 29,  // Next Line
+  FX_CBP_SA = 30,  // Complex Context (South East Asian)
+  FX_CBP_SG = 31,  // Surrogate
+  FX_CBP_CB = 32,  // Contingent Break Opportunity
+  FX_CBP_XX = 33,  // Unknown
+  FX_CBP_AI = 34,  // Ambiguous (Alphabetic or Ideographic)
+  FX_CBP_SP = 35,  // Space
   FX_CBP_NONE = 36,
+  FX_CBP_TB = 37,  // ?
 };
 
 #define FX_CHARTYPEBITS 11
@@ -112,7 +114,6 @@ inline FX_CHARTYPE GetCharTypeFromProp(uint32_t prop) {
   return static_cast<FX_CHARTYPE>(prop & FX_CHARTYPEBITSMASK);
 }
 
-bool FX_IsCtrlCode(FX_WCHAR ch);
 FX_WCHAR FX_GetMirrorChar(FX_WCHAR wch,
                           uint32_t dwProps,
                           bool bRTL,
diff --git a/core/fxcrt/fx_unicode.cpp b/core/fxcrt/fx_unicode.cpp
index e6f423e2af..44fa8e743d 100644
--- a/core/fxcrt/fx_unicode.cpp
+++ b/core/fxcrt/fx_unicode.cpp
@@ -13,13 +13,6 @@ uint32_t FX_GetUnicodeProperties(FX_WCHAR wch) {
   return 0;
 }
 
-#ifdef PDF_ENABLE_XFA
-bool FX_IsCtrlCode(FX_WCHAR ch) {
-  uint32_t dwRet = (FX_GetUnicodeProperties(ch) & FX_CHARTYPEBITSMASK);
-  return dwRet == FX_CHARTYPE_Tab || dwRet == FX_CHARTYPE_Control;
-}
-#endif  // PDF_ENABLE_XFA
-
 FX_WCHAR FX_GetMirrorChar(FX_WCHAR wch, bool bRTL, bool bVertical) {
   uint32_t dwProps = FX_GetUnicodeProperties(wch);
   uint32_t dwTemp = (dwProps & 0xFF800000);
diff --git a/xfa/fgas/layout/fgas_rtfbreak.cpp b/xfa/fgas/layout/fgas_rtfbreak.cpp
index d9e5f7f80b..8f5895f24c 100644
--- a/xfa/fgas/layout/fgas_rtfbreak.cpp
+++ b/xfa/fgas/layout/fgas_rtfbreak.cpp
@@ -9,7 +9,6 @@
 #include <algorithm>
 
 #include "core/fxcrt/fx_arabic.h"
-#include "core/fxcrt/fx_arb.h"
 #include "third_party/base/stl_util.h"
 #include "xfa/fgas/font/cfgas_gefont.h"
 #include "xfa/fgas/layout/fgas_linebreak.h"
@@ -470,7 +469,7 @@ bool CFX_RTFBreak::EndBreak_SplitLine(CFX_RTFLine* pNextLine,
     }
   }
 
-  if (!m_bPagination && m_pCurLine->m_iMBCSChars <= 0) {
+  if (!m_bPagination) {
     if (bAllChars && !bDone) {
       int32_t endPos = m_pCurLine->GetLineEnd();
       GetBreakPos(m_pCurLine->m_LineChars, endPos, bAllChars, true);
@@ -1013,11 +1012,7 @@ CFX_RTFPiece::~CFX_RTFPiece() {
 }
 
 CFX_RTFLine::CFX_RTFLine()
-    : m_LinePieces(16),
-      m_iStart(0),
-      m_iWidth(0),
-      m_iArabicChars(0),
-      m_iMBCSChars(0) {}
+    : m_LinePieces(16), m_iStart(0), m_iWidth(0), m_iArabicChars(0) {}
 
 CFX_RTFLine::~CFX_RTFLine() {
   RemoveAll(false);
diff --git a/xfa/fgas/layout/fgas_rtfbreak.h b/xfa/fgas/layout/fgas_rtfbreak.h
index 63ad67f6c0..8907871f74 100644
--- a/xfa/fgas/layout/fgas_rtfbreak.h
+++ b/xfa/fgas/layout/fgas_rtfbreak.h
@@ -127,7 +127,6 @@ class CFX_RTFLine {
     m_LinePieces.RemoveAll(bLeaveMemory);
     m_iWidth = 0;
     m_iArabicChars = 0;
-    m_iMBCSChars = 0;
   }
 
   std::vector<CFX_RTFChar> m_LineChars;
@@ -135,7 +134,6 @@ class CFX_RTFLine {
   int32_t m_iStart;
   int32_t m_iWidth;
   int32_t m_iArabicChars;
-  int32_t m_iMBCSChars;
 };
 
 class CFX_RTFBreak {
diff --git a/xfa/fgas/layout/fgas_textbreak.cpp b/xfa/fgas/layout/fgas_textbreak.cpp
index 8be72f2c7b..153d575139 100644
--- a/xfa/fgas/layout/fgas_textbreak.cpp
+++ b/xfa/fgas/layout/fgas_textbreak.cpp
@@ -9,7 +9,6 @@
 #include <algorithm>
 
 #include "core/fxcrt/fx_arabic.h"
-#include "core/fxcrt/fx_arb.h"
 #include "core/fxcrt/fx_memory.h"
 #include "third_party/base/ptr_util.h"
 #include "xfa/fgas/font/cfgas_gefont.h"
@@ -31,6 +30,11 @@ const FX_TxtBreak_LPFAppendChar g_FX_TxtBreak_lpfAppendChar[16] = {
     &CFX_TxtBreak::AppendChar_Others,      &CFX_TxtBreak::AppendChar_Others,
 };
 
+bool IsCtrlCode(FX_WCHAR ch) {
+  uint32_t dwRet = (FX_GetUnicodeProperties(ch) & FX_CHARTYPEBITSMASK);
+  return dwRet == FX_CHARTYPE_Tab || dwRet == FX_CHARTYPE_Control;
+}
+
 }  // namespace
 
 CFX_TxtBreak::CFX_TxtBreak(uint32_t dwPolicies)
@@ -1561,7 +1565,7 @@ std::vector<CFX_RectF> CFX_TxtBreak::GetCharRects(const FX_TXTRUN* pTxtRun,
       iCharSize = *pWidths++;
     }
     fCharSize = static_cast<FX_FLOAT>(iCharSize) / 20000.0f;
-    bool bRet = (!bSingleLine && FX_IsCtrlCode(wch));
+    bool bRet = (!bSingleLine && IsCtrlCode(wch));
     if (!(wch == L'\v' || wch == L'\f' || wch == 0x2028 || wch == 0x2029 ||
           (wLineBreakChar != 0xFEFF && wch == wLineBreakChar))) {
       bRet = false;